China’s approach to AI governance is entering a new phase. Having spent the past three years building the world’s largest AI computing infrastructure and deploying models across virtually every sector of the economy, the country’s regulators are now turning their attention to a more fundamental question: how do you measure whether AI systems are actually working as intended? On May 28, the State Administration for Market Regulation (SAMR) and the National Development and Reform Commission (NDRC) jointly released a guideline for developing a national AI metrology system, according to Xinhua.
Metrology, the science of measurement, is the unglamorous infrastructure that underpins every advanced technology sector. In semiconductors, metrology determines whether a chip meets its specifications. In pharmaceuticals, it ensures that a drug contains what it claims to contain. In AI, metrology addresses questions such as: Does this model perform as accurately as its benchmark scores suggest? Does it behave consistently across different deployment environments? Can its outputs be trusted for high-stakes decisions in healthcare, finance, or infrastructure?
Six Areas of Focus
The guideline organizes China’s AI metrology development around six key areas. The first is foundational support, the basic measurement infrastructure, standards bodies, and laboratory capabilities required to assess AI systems. The second is general technology, covering the measurement methods applicable across different types of AI. The third is core technology, addressing the specific measurement challenges posed by large language models, computer vision systems, and other leading AI architectures. The fourth area covers metrological technical standards, the formal standards that will define what “accurate” and “reliable” mean for AI systems in regulated contexts. The fifth is the metrology service industry, the commercial ecosystem of testing, certification, and calibration providers that will implement the standards in practice. The sixth is the intelligent empowerment of metrology itself, using AI to improve measurement science.
SAMR described the guideline as marking a critical shift from expanding computing power and scale to improving quality and strengthening foundational capabilities. The framing is significant: it acknowledges that China’s AI industry has reached a scale at which raw capacity is no longer the binding constraint, and that the next phase of value creation requires quality assurance infrastructure that the industry currently lacks.
The 15th Five-Year Plan Connection
The metrology guideline is explicitly linked to China’s 15th Five-Year Plan (2026–2030), released in March 2026, which emphasizes comprehensive implementation of the “AI Plus” initiative, the policy framework for integrating AI across all sectors of the economy. SAMR stated that the guideline is of great significance for promoting the deep integration of AI technology with the real economy and accelerating the development of new-quality productive forces, a term used in Chinese policy discourse to describe AI-driven productivity gains.
The connection to the Five-Year Plan matters because it signals that AI metrology is not a niche regulatory concern but a national priority with dedicated resources and political backing. As EastFrontier has reported, China’s AI regulation entered a new phase and China has accelerated AI and low-altitude economy legislation to support the Five-Year Plan. The metrology roadmap adds a technical measurement layer to what has until now been primarily a legal and regulatory framework.
For international AI companies operating in China, the metrology guideline has practical implications. As measurement standards are formalized, AI systems deployed in regulated sectors will need to demonstrate compliance, creating both a compliance burden and a competitive barrier for foreign providers who may not have designed their systems with Chinese measurement standards in mind.
The international dimension of the guideline extends beyond market access. Metrology standards, once established, tend to propagate through trade relationships: countries that import Chinese AI products or technology may find themselves adopting Chinese measurement frameworks as a condition of interoperability. This is the same dynamic that has played out in 5G (where Huawei’s equipment set de facto standards in many markets) and in electric vehicle charging (where China’s GB/T standard has been adopted across much of Asia and Africa). If China moves quickly to establish AI metrology standards and exports them through its Belt and Road technology partnerships, it gains the ability to shape how AI performance is measured and certified globally, a form of soft power that operates through technical infrastructure rather than diplomatic negotiation. The metrology guideline is, in this sense, not just a domestic quality-assurance document but a strategic instrument in the global AI standards competition.